php 在cli模式下使用curl模拟登录保存cookie，爬取

<?php


error_reporting(E_ALL);
ini_set('display_errors', 1);

//浏览器友好输出
function dump($var)
{
    ob_start();
    var_dump($var);
    $output = ob_get_clean();

    $output = preg_replace('/\]\=\>\n(\s+)/m', '] => ', $output);

    if (PHP_SAPI !== 'cli') {
        if (!extension_loaded('xdebug')) {
            $output = htmlspecialchars($output, ENT_HTML5 | ENT_SUBSTITUTE, 'UTF-8');
        }
        $output = '<pre>' . $output . '</pre>';
    } else {
        $output = PHP_EOL . $output;
    }

    echo $output;
    //exit();
}


function http_curl($url, $data = null, $method = 'get', $token = '', $cookieName = '')
{

    if (is_array($data)) {
        $data = json_encode($data, 320);
    }

    $ch = curl_init(); // 初始化 cURL 资源

    // 设置 URL
    curl_setopt($ch, CURLOPT_URL, $url);

    // 设置请求方式和数据
    if ($method == 'post') {
        curl_setopt($ch, CURLOPT_POST, true);
        curl_setopt($ch, CURLOPT_POSTFIELDS, $data);
    }

    // 设置请求头部信息
    curl_setopt($ch, CURLOPT_HTTPHEADER, array(
        'Content-Type: application/json; charset=utf-8',
        'Content-Length:' . strlen($data),
        'token: ' . $token // 请求头携带的token
    ));

    // 设置是否将结果保存到变量而不是直接输出
    curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);

    // 设置是否跳过证书验证
    curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
    curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, false);

    // 设置连接和读取超时时间
    curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 5);
    curl_setopt($ch, CURLOPT_TIMEOUT, 10);

    // 设置保存和读取 Cookie
    if ($cookieName) {
        $cookie_file = dirname(__FILE__) . '/' . $cookieName . '_cookie.txt';
        curl_setopt($ch, CURLOPT_COOKIEJAR, $cookie_file);
        curl_setopt($ch, CURLOPT_COOKIEFILE, $cookie_file);
    }

    $result = curl_exec($ch); // 执行请求

    curl_close($ch); // 关闭 cURL 资源

    return json_decode($result, true);
}

function simulateLogin()
{
    //获取用户名和密码
    echo "请输入用户名：";
    $username = trim(fgets(STDIN));
    var_dump($username) . "\n";

    echo "请输入密码：";
    $password = trim(fgets(STDIN));
    var_dump($password) . "\n";

    // 获取验证码
    // 这里假设验证码是一个四位数
    // echo "请输入验证码（四位数）：";
    // $captcha = trim(fgets(STDIN));
    // echo $captcha."\n";

    $login_url = "https://www.test.com/api/login";

    $data = [
        'phone' => $username,
        'password' => $password
    ];
    $res = http_curl($login_url, $data, 'post');

    if ($res && $res['code'] == 0) {
        echo "登录成功！" . "\n";
        echo "抓取数据，请稍等..." . "\n";
        //抓取数据
        $data = get_data($res['data']['token']);
        dump($data);
        //循环插入数据库动作

    } else {
        var_dump($res['data']);
    }
}

//爬取数据
function get_data($token)
{
    //请求url
    $url = "https://www.test.com/api/dt/rlist";
    //$token = 'eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9374k0wGubr8ZThq_svNDJg';
    //第一次请求获取第一页数据和总页码
    $res = http_curl($url, '', 'post', $token);
    $last_page = $res['data']['last_page'];//获取总页数
    $allData = $res['data']['data'];//获取的第一页数据
    //获取到总页数后，从第二页循环获取每页数据
    for ($page = 2; $page <= $last_page; ++$page) {
        $data = [
            'page' => $page
        ];
        $res = http_curl($url, $data, 'post', $token);
        //合并数据
        $allData = array_merge($allData, $res['data']['data']);
        //dump($allData);

        // 插入数据库的动作
        //foreach ($allData as $item) {
        // 可以将$item中的数据插入到数据库中
        // insertDataToDatabase($item);
        //}
    }
    return $allData;
}

//调用
simulateLogin();
?>